Method and system for visually detecting deviation of training action posture of competitive sports athletes
By reconstructing 3D joint coordinates using binocular cameras and deep learning, and combining closed-chain biomechanical constraints and phase-adaptive temporal registration, the problem of high-precision detection of athlete movement posture deviations in existing technologies has been solved. This enables high-precision, low-false-alarm detection in unmarked deployments on training grounds, accurately pinpointing root causes and improving training effectiveness.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHONGQING CITY VOCATIONAL COLLEGE
- Filing Date
- 2026-04-23
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies cannot accurately detect athletes' movement and posture deviations in competitive sports training with high precision and without marking, nor can they accurately pinpoint the root causes, making it impossible for coaches to effectively correct these deviations during training.
A high-speed binocular camera is used to capture videos of athletes. A deep learning human pose estimation network is used to detect joint points and reconstruct three-dimensional joint coordinates. The closed-chain biomechanical constraint envelope is extracted from a variety of excellent samples. Combined with phase adaptive temporal registration and kinematic chain back tracing, high-precision, low-false-alarm motion posture deviation detection is achieved.
It enables high-precision detection of athletes' movement and posture deviations in unmarked training environments, accurately pinpointing the root causes and improving detection specificity and the effectiveness of training interventions.
Smart Images
Figure CN122415535A_ABST